Proc. Nati. Acad. Sci. USA Vol. 91, pp. 3593-3597, April 1994 Biochemistry A recombinant bisphosphoglycerate mutase variant with acid phosphatase homology degrades 2,3-diphosphoglycerate MARIE-CLAUDE GAREL, NICOLE AROUS, MARIE-CLAUDE CALVIN, CONSTANTIN TIGELIU CRAESCU, JEAN ROSA, AND RAYMONDE ROSA Institut National de la Sante et de la Recherche M6dicale, U.91, H6pital Henri Mondor, 94010 Cr6teil, France Communicated by M. F. Perutz, January 3, 1994 (receivedfor review September 6, 1993)

ABSTRACT To date no definite and undisputed treatment His in order to facilitate the orientation and binding of the has been found for sickle cell anemia, which is characterized by phosphate group and consequently activate the phosphoryl polymerization of a deoxygenated hemoglobin mutant (HbS) transfer to water (2). giving rise to deformed erythrocytes and vasoocclusive com- Little is known about the three-dimensional structure and plications. Since the erythrocyte glycerate 2,3-bisphosphate particularly localization of the active site residues in acid (2,3-DPG) has been shown to facilitate this polymerization, one phosphatases. Conversely, the structure of the yeast PGM therapeutic approach would be to decrease the intraerythro- has been determined by x-ray diffraction analysis by using cytic level of2,3-DPG by increasing the phosphatase activity of crystals soaked in 3-phosphoglycerate (3-PG). This latter the bisphosphoglycerate mutase (BPGM; 3-phospho-D- enzyme shares 5Ow sequence identity with human BPGM glycerate 1,2-phosphomutase, EC 5.4.2.4). For this purpose, and catalyzes the same three reactions (synthase, mutase, we have investigated the role of Gly-13, which is located in the and phosphatase), although at substantially different rates. active site sequence Arg'-His"'-Gly"1-Glu12-Gly13 in human These activities are catalyzed at the unique active site of BPGM. This sequence is similar to the Arg-His-Gly-Xaa-Arg* BPGM. To date and in spite of recent crystallization of sequence of the distantly related acid phosphatases, which human BPGM (9), no crystallographic data have been ob- catalyze as BPGM similar phosphoryl transfers but to a greater tained for this enzyme. Consequently, the amino acids in- extent. We hypothesized that the conserved Arg* residue in volved in the active site ofhuman BPGM have been deduced acid phosphatase sequences facilitates the phosphoryl transfer. by comparison with the structure of the yeast PGM (8). Consequently, in human BPGM, we replaced by site-directed Amino acid residues of the active site were highly conserved mutagenesis the corresponding amino acid residue Gly'3 with between BPGM and PGM. The different catalytic rates an Arg or a Lys. In another experiment, we replaced Gly'3 with observed for these two homologous enzymes could be ex- Ser, the amino acid present at the corresponding position ofthe plained by the nonconserved residues in their active site. homologous yeast phosphoglycerate mutase (n-phosphoglycer- Among them, Gly13 in BPGM and the homologous Ser11 in the ate 2,3-phosphomutase, EC 5.4.2.1). Mutation of Gly'3 to Ser yeast PGM have been postulated to play an important role (7, did not modify the synthase activity, whereas the mutase and 8). the phosphatase were 2-fold increased or decreased, respec- Synthase and phosphatase activities of BPGM catalyze, tively. However, replacing Gly'3 with Arg enhanced phospha- respectively, the synthesis and degradation of glycerate tase activity 28.6-fold, whereas synthase and mutase activities 2,3-bisphosphate (2,3-DPG), the main allosteric effector of were 10-fold decreased. The presence of a Lys in position 13 hemoglobin. In spite of the presence of acid phosphatases in gave rise to a smaller increase in phosphatase activity (6.5-fold) erythrocytes, the degradation of2,3-DPG is very low because but an identical decrease in synthase and mutase activities. these acid phosphatases cannot degrade 2,3-DPG and the Taken together these results support the hypothesis that a phosphatase activity of BPGM is very slow (1000-fold lower positively charged amino acid residue in position 13, especially than the synthase activity and lower than that of acid phos- Arg, greatly activates the phosphoryl transfer to water. These phatases). Consequently, 2,3-DPG is present in high concen- results also provide elements for locating the conserved Arg* trations in erythrocytes. residue in the active site of acid phosphatases and facilitating Our purpose being to decrease 2,3-DPG level in erythro- the phosphoryl transfer. The implications for genetic therapy cytes by increasing the BPGM phosphatase activity, we have of sickle cell disease are discussed. compared amino acid sequences ofseveral acid phosphatases possessing a His residue in their active site and especially the Some enzymes displaying phosphatase activities such as sequences of a fragment common to these enzymes and human prostatic acid phosphatase (1, 2), lysosomal (3) and human BPGM (Fig. 1) (4). This fragment, corresponding to yeast (4) acid phosphatases, erythrocyte bisphosphoglycer- the amino acid sequence Arg9-His10-Gly11-Glu12-Gly13 in hu- ate mutase (BPGM; 3-phospho-D-glycerate 1,2-phosphomu- man BPGM, is homologous to the amino acid sequence tase, EC 5.4.2.4) (5-7), glycolytic phosphoglycerate mutase Arg7-His8-Gly9-Gln10-Ser'1 in yeast PGM, which is localized (PGM; D-phosphoglycerate 2,3-phosphomutase, EC 5.4.2.1) in the active site and contains the phosphorylatable His (7, 8, (7, 8), and hepatic 6-phosphofructo-2-kinase/fructose-2,6- 10). From this analysis, we have postulated that the con- bisphosphatase (fructose-2,6-bisphosphate 2-phosphatase; served Arg residue, denoted Arg* in the acid phosphatase D-fructose-2,6-bisphosphate 2-phosphohydrolase, EC consensus sequence and which corresponds to the Gly13 in 3.1.3.46) (1) possess a His residue in their active site, which BPGM, could be a good candidate as a cationic group for is transiently phosphorylated during the course of the phos- enhancing the phosphoryl transfer. We have therefore sub- phoryl transfer. It was suggested that most such enzymes stituted Arg for Gly13 in human BPGM by site-directed involved in the binding of phosphate esters should have at mutagenesis. We have also substituted Lys for Gly13 to least one cationic group suitably disposed near the active site evaluate the role of another positively charged amino acid at

The publication costs of this article were defrayed in part by page charge Abbreviations: 2,3-DPG, glycerate 2,3-bisphosphate; 3-PG, glycer- payment. This article must therefore be hereby marked "advertisement" ate 3-phosphate; BPGM, bisphosphoglycerate mutase; PGM, phos- in accordance with 18 U.S.C. §1734 solely to indicate this fact. phoglycerate mutase. 3593 Downloaded by guest on October 2, 2021 3594 Biochemistry: Garel et al. Proc. NatL. Acad. Sci. USA 91 (1994) Hu-BPGM Arg9 His10 GIyll G1u12 G1y13 tion was performed on an HPLC column of Fractogel TSK AF blue (Merck) at room temperature as reported (17). HU-PGM-M Arg His Gly G1u Thr Enzyme and 2,3-DPG Assays. Synthase and mutase activ- Ye-PGM Arg7 His8 GIy9 Gin10 Ser11 ities were assayed according to methods already reported (18). When synthase activity was assayed directly on the E. F-2,6-P2ase Arg His Gly Glu Ser colicrude extracts, this assay being directly related to NADH production by glyceraldehyde phosphate dehydrogenase, the reaction could be partially masked by the intrinsic NADH Hu-P-ACP Arg His Gly Asp Arg* oxidase activity ofE. coli. It was, therefore, necessary to add 2 mg ofantimycin A and 1 mmol ofKCN to 1 ml ofthe assay Hu-L-ACP Arg His GIl ASP Arg* system as specific inhibitors of E. coli NADH oxidase. The Ye-ACP1 Arg His Gly Ser Arg* phosphatase activity was measured by coupling the reaction with phosphoglycerate kinase, glyceraldehyde phosphate Ye-ACP3 Arg His Gly Glu Arg* dehydrogenase, triosephosphate isomerase, and glycerol Ye-ACP5 Arg His Gly GIu Arg* phosphate dehydrogenase. In this sequence of reactions, the degradation of 1 mol of 3-PG is coupled with the oxidation of FIG. 1. Conserved peptide sequence in human BPGM (Hu- 2 mol of NADH. Each sample was checked against a control BPGM), human muscle PGM (Hu-PGM-M), yeast PGM (Ye-PGM), deprived of 2,3-DPG. In the case of 2-phosphoglycolate hepatic 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase (F- stimulation of the phosphatase reaction, 1 mM 2-phospho- 2,6-P2ase), human prostatic acid phosphatase (Hu-P-ACP), human glycolate was added to the standard assay system and be- lysosomal acid phosphatase (Hu-L-ACP), yeast acid phosphatase P1 cause of its potent effect (19) incubation was performed for (Ye-ACP1), yeast acid phosphatase P3 (Ye-ACP3), and yeast acid 15 min instead of 60 min and 0.4 mg of the enzyme was used phosphatase PS (Ye-ACP5). instead of 4 mg. The amount of 2,3-DPG was measured this position in the active site. Another substitution was enzymatically on the deproteinized extracts by techniques performed by replacing Gly" with Ser to analyze the modi- previously reported (18), and its amount was related to the fications ofthe catalytic properties of BPGM when this Gly13 ratio ofprotein in the lysate measured according to Lowry et is replaced by the amino acid residue present in the homol- al. (20). ogous yeast PGM. Electrophoresis. PAGE was performed in the presence of SDS according to Laemmli (21). The gels were stained with Coomassie blue R250 and the amount of expressed wild-type MATERIALS AND METHODS or mutant BPGM was determined by densitometric scanning Materials. Except when specified otherwise the reagents of the gels. used for the buffers were obtained from Merck. All substrates Thermostability Studies. The purified variants were incu- and commercial enzymes were purchased from Boehringer bated at 550C for 30 min in 10 mM Tris HCl buffer (pH 7.5) Mannheim except for NADH, which was a product of Sigma containing 1 mM EDTA, 1 mM 2-mercaptoethanol, and 1 mg as were Trizma base (Tris), bovine serum albumin, and of bovine serum albumin per ml. The incubation was stopped dithiothreitol. Acrylamide and bisacrylamide were supplied by placing the tubes in ice water. After centrifugation, by Fluka. Isopropyl alcohol was obtained from Prolabo antibody consumption was determined as previously re- (Paris). Rabbit antiserum directed against human erythrocyte ported (22). BPGM was obtained according to methods previously re- Modeling of the Enzymes. Molecular modeling of the en- ported (11). Oligonucleotides were synthesized by the phos- zymes was performed using the UNIX version of the BRUGEL phoramidite method on an Applied Biosystems DNA syn- software package as described (10). Modeling of the three- thesizer (model 381A) followed by purification by electro- dimensional conformation of the human BPGM enzyme was phoresis on a 20%6 polyacrylamide gel containing 7 M urea. made by using the known crystallographic structure of the Site-Directed Mutagenesis. The procedure used for oligo- homologous yeast PGM as a starting structure (8). nucleotide-directed site-specific mutagenesis was based on the method described by Taylor et al. (12) using a kit RESULTS developed by Amersham and 20-mer oligonucleotides encod- at Expression of the Mutated BPGM Enzymes. The site- ing single amino acid mutations residue 13. At the end of directed mutagenesis experiments were performed as de- thetheprocedure,procedure,sngle-strandedsingle-strandedunAs ofofputativemutputative mutant scribed. For all the BPGM mutants, the entire coding region phages were prepared and sequenced by the dideoxynucle- of the cDNA was sequenced by using specific internal otide cham-termination method (13) to confirm the desired oligonucleotides in order to confirm that no additional mu- mutation. The complete sequence of the mutated BPGM tation was introduced during mutagenesis. insert was then checked entirely between the restriction sites Large amounts of the wild type and the three mutants of used for subcloning in the expression vector pKK223-3 human BPGM (Gly13 to Arg, Gly13 to Lys, and Gly13 to Ser) (Pharmacia) as described (14). For all these procedures, the were expressed in E. coli by using the expression vector basic cloning methods described by Maniatis et al. were used pKK223-3 as described for the wild-type human BPGM (14). (15). Each BPGM variant and the wild-type enzyme were pro- Preparation of Overexpressed Mutants and Wild-Type Hu- duced in E. coli in the same amount (-5.5% of the bacterial man BPGM. Overnight cultures of Escherichia coli contain- protein). As demonstrated for the wild-type enzyme, the ing an expression plasmid of either mutant or wild-type growth rate of bacteria was decreased every time 2,3-DPG enzyme were grown in L broth medium containing ampicillin was synthesized by the recombinant enzyme (14). Such an (100 ug/ml) and then diluted 1:100 for the expression cul- inhibition of the growth rate of bacteria was not observed tures. Induction of protein expression was performed with during expression of the Gly13 to Arg and Gly13 to Lys isopropyl /3-D-thiogalactoside as described (14). variants, which produced very low levels of 2,3-DPG as Purification of the mutant and wild-type enzymes was described below. In contrast, this inhibition was similar to performed as described (16) except that the lysate was not that of the wild-type enzyme when the Gly13 to Ser variant, heated before chromatography and the first step of purifica- which could synthesize 2,3-DPG, is expressed. Downloaded by guest on October 2, 2021 Biochemistry: Garel et al. Proc. Natl. Acad. Sci. USA 91 (1994) 3595

Catalytic Properties of Gly'3 Variants. The catalytic prop- erties of the three variants are summarized in Table 1 and compared to those of normal human BPGM expressed in E. coli. When Gly'3 was replaced by Ser, a slight decrease (2-fold) of the phosphatase activity was observed and there was a lower capacity of its stimulation by 2-phosphoglyco- late. Simultaneously the synthase activity was normal and the is mutase activity was 2-fold increased. By contrast, when Arg or Lys replaced Glyl3 a very strong increase in the phospha- tase activity (28.6- and 6.5-fold, respectively) was observed, whereas phosphatase stimulation by 2-phosphoglycolate was decreased. These two variants (Glyl3 to Arg and Glyl3 to Lys) showed decreased synthase (10- and 8.6-fold, respectively) and mutase (10- and 6.4-fold, respectively) activities. It should be noted that the control values of the phosphatase activities and Km values were different from those previously reported (16) because of the differences in the two assay techniques. The Michaelis constants for the substrates in the three a b c d e reactions catalyzed by wild-type and mutant human BPGM were measured (16) and are summarized in Table 2. The Km FIG. 2. Coomassie blue-stained SDS/15% polyacrylamide gel of for 3-PG in the synthase activity and for 2-PG in the mutase purified BPGM Gly13 to Arg variant used in this study. Lanes: a, activity were slightly modified (2-fold). molecular mass markers (from top) of 94, 67, 43, 30, and 14.4 kDa; The Michaelis constants for the binding of 2,3-DPG were b, purified wild-type human BPGM expressed in E. coli; c, total cell extract ofE. coli expressing the Gly13 to Arg variant after induction; 2- to 6.7-fold decreased in the basic phosphatase activity and d, partially purified Gly13 to Arg variant after the first blue Fractogel the mutase activity of the three variants except for a slight column chromatography; e, purified Gly13 to Arg variant after the increase in the mutase activity for the Gly'3 to Arg variant. second step ofpurification on a Fractogel TSK column. The mutated In contrast, a moderate increase was observed for the Gly13 oligonucleotide for production of the Gly13 to Arg variant by site- to Arg and Gly13 to Lys variants in the 2-phosphoglycolate- directed mutagenesis according to the method described by Taylor et stimulated phosphatase activity (3.5- and 2.5-fold, respec- al (12) has the sequence 5'-CATGGAGAGCGTGCTTGAATT-3'. Km the in the synthase activity Construction of the expression vector, expression of the wild-type tively). The value for 1,3-DPG and mutant BPGM, and their purification have been described (14, could not be determined because of the great instability of 16). Induction ofthe tac promoter ofthe expression vector was made this substrate. with 0.5 mM isopropyl f-D-thiogalactoside during 3 h at 37°C. DISCUSSION Neither synthase activity nor detectable amounts of 2,3- DPG could be found in crude extracts of nontransformed E. In this report, we show that replacement by site-directed coli XL1-B bacteria. Consequently, we measured these val- mutagenesis of Glyl3 by Ser in human BPGM did not greatly ues in the lysate of the bacteria expressing the different modify the catalytic properties of the mutant enzyme as recombinant enzymes. When bacteria expressed the Gly13 to compared to the wild-type BPGM. The slightly decreased Arg and Gly13 to Lys variants, we detected a low synthase phosphatase activities as well as increased mutase activity activity (0.038 and 0.11 unit per mg of protein, respectively) mimic the PGM enzyme, and these results are in accord with as well as a low 2,3-DPG production (0.0016 and 0.0015 umol those obtained for the corresponding yeast PGM Ser" to Gly of 2,3-DPG per mg of protein, respectively). In contrast, a variant recently produced by White and Fothergill-Gilmore synthase activity (0.41 unit per mg of protein) similar to that (23) (as shown in Fig. 1, the BPGM Glyl3 residue corresponds obtained with the wild-type enzyme (0.45 unit per mg of to the yeast PGM Ser" residue). X-ray diffraction analysis of protein) was present in the crude extracts of E. coli trans- yeast PGM crystals soaked in 3-PG has shown that the Ser" formed with the plasmid bearing the Gly13 to Ser mutation, residue could directly interact with substrate (8). Our results giving rise to production of 0.027 ,umol of 2,3-DPG per mg of of Km for the Gly13 to Ser BPGM variant show that the protein (control, 0.054 umol of 2,3-DPG per mg of protein). monophosphoglycerate affinities are only slightly modified All the variants were purified to homogeneity (14, 16). A while the affinity for 2,3-DPG increases. Such results are single polypeptide band with a molecular mass of 30 kDa was comparable to those obtained for the Ser" to Gly PGM obtained by SDS/PAGE for each purified enzyme after variant, which showed a greatly decreased (10-fold) affinity Coomassie blue staining as shown in Fig. 2 for the Gly13 to for 2,3-DPG. These results afford evidence for interactions Arg variant. All the variants were stable when incubated at between 2,3-DPG and the serine residue located at a homol- 55°C for 30 min. ogous position in the active sites of the human BPGM and Table 1. Specific activities for purified wild-type, Gly13 to Arg, Gly13 to Lys, and Gly13 to Ser variants of human BPGM expressed in E. coli Specific activity, units per mg of protein Enzyme Reaction Wild type Gly13 to Arg Gly13 to Lys Gly13 to Ser Synthase 1,3-DPG - 2,3-DPG 16.3 1.5 1.9 15.8 Mutase 2-PG 3-PG 8.5 0.8 1.3 17.3 Stimulated phosphatase 2,3-DPG 3-PG+Pi 7.6 1.3 0.3 4.7 Unstimulated phosphatase 2,3-DPG - 3-PG+Pi 2.45 x 10-2 70.2 x 10-2 15.9 x 10-2 1.34 x 10-2 Downloaded by guest on October 2, 2021 35% Biochemistry: Garel et al. Proc. Natl. Acad. Sci. USA 91 (1994) Table 2. Km values for substrates of the three reactions catalyzed by purified wild-type, Gly13 to Arg, Gly13 to Lys, and Gly13 to Ser variants of human BPGM expressed in E. coli Ki, mM Enzyme Wild type Gly13 to Arg Gly13 to Lys Gly13 to Ser Synthase 3-PG 80 35 52 71 Mutase 2-PG 15 8.6 29 18.5 2,3-DPG 125 416 65 25 Unstimulated phosphatase 2,3-DPG 3570 680 530 1040 Stimulated phosphatase 2,3-DPG 41 142 104 25 Km values were determined according to the methods previously reported (16) except for the phosphatase activities, whose determination is described in Materials and Methods. Calculations were from Lineweaver-Burke plots. Measurements are averages of at least three determinations.

yeast PGM enzymes and suggest that this residue is not phatase reaction inhibited the binding of2,3-DPG as revealed involved in the binding of monophosphoglycerates. by the increased Km values. In contrast, as postulated, replacement of Gly'3 in human The Gly'3 to Arg mutation in human BPGM greatly in- BPGM by a positively charged amino acid residue greatly creasing the phosphatase activity suggests that the guanidium modifies the relative ratios ofthe three reactions. For the two group ofArgl3 stabilizes the transition state ofthe transferred Glyl3 to Arg and Gly13 to Lys BPGM variants, the synthase phosphate, promoting its transferfrom His10 to water. So, the and mutase reactions are greatly decreased. On the contrary, essential feature ofthe proposed mechanism ofaction ofacid the basic phosphatase activity is markedly increased for the phosphatases using a phosphohistidine intermediate is ex- Glyl3 to Arg variant and to a lesser extent by the Gly'3 to Lys perimentally demonstrated in this paper with human BPGM. variant. Such results show that in human BPGM, an Arg The structure of human BPGM was modeled (10) using residue present at position 13 strongly increases the phos- crystallographic data of the well-known homologous yeast phoryl transfer to a water molecule. Nevertheless, the pres- PGM and amino acid residues involved in the active site have ence in the active site of a monophosphoglycerate (3-PG) been proposed. Some ofthem are represented in Fig. 3. It can during the mutase reaction or 2-phosphoglycolate in the be seen that the Glyl3 residue is located in the neighborhood inhibited this phosphoryl of the phosphorylatable His10 residue. Replacement of this stimulated phosphatase reaction, Gly'3 by Arg followed by energy minimization (data not transfer. In this paper, the Km values for monophosphoglyc- shown) indicates that the new guanidium group could point erates and 2,3-DPG show that their affinities are slightly toward the active site and occupies a position located in the (2-fold) modified. Nevertheless, for the Gly13 to Arg and neighborhood ofHis'0 and Cys'2 residues. In such a position, Gly13 to Lys variants, it is remarkable that the 2,3-DPG it can be postulated that the new guanidium group could affinities are greatly increased in the phosphatase activity, interact with the substrates. indicating that the presence of Arg3 or Lys'3 residues led to In conclusion, if the peptide that has a sequence similar to better interactions with 2,3-DPG. However, the presence of that of BPGM performs a similar function in all the acid 2-phosphoglycolate in the active site in the stimulated phos- phosphatases, it is reasonable to assume that, as in BPGM, it contains the active site (phospho)His residue and that the H 187 conserved Arg* residue is located in their active site. In addition, it is likely that activation ofthe phosphoryl transfer to a water molecule obtained in this report for human BPGM with the Gly'3 to Arg mutation is not unique to this enzyme but represents a common mechanism for other phosphatases possessing His residues in their active site. It would be interesting to make the corresponding mutation in the yeast PGM (7) and fructose-2,6-bisphosphatase (1) to show that such a mutation also activates the phosphatase activity. Only in vitro modulation of 2,3-DPG level has been per- formed to study its effect on polymerization of deoxyHbS (24). The recent production by several laboratories (25-27) and ourselves (28) of a transgenic mouse model of sickle cell disease (29-33) offers the opportunity to evaluate the poten- tial role of a decreased 2,3-DPG level in vivo on erythrocyte sickling by expressing the Gly'3 to Arg BPGM variant in their sickle erythrocytes. We thank Dr. N. Blumenfeld for assistance in revision of the English. We are grateful to A. M. Dulac, R. Quintel, andJ. M. Masse for preparation of the manuscript and the figures. This work was supported by grants from the Institut National de la Sante et de la Recherche Medicale and the Centre National de la Recherche Scientifique. FIG. 3. View of the active site of human BPGM modeled on the basis of crystallographic coordinates of PGM (8, 10). Only the 1. Bazan, J. F., Fletterick, R. J. & Pilkis, S. J. (1989) Proc. Natl. residues considered to play a major role in ligand binding are shown. Acad. Sci. USA 86, 9642-9646. Downloaded by guest on October 2, 2021 Biochemistry: Garel et al. Proc. Natl. Acad. Sci. USA 91 (1994) 3597 2. Van Etten, R. L. (1982) Ann. N. Y. Acad. Sci. 390, 27-51. Craescu, C. T., Prehu, M.-O., Rosa, J. & Rosa, R. (1993) Eur. 3. Pohlmann, R., Krentler, C., Schmidt, B., Schr6der, W., J. Biochem. 213, 493-500. Lorkowski, G., Culley, J., Mersmann, G., Geier, C., Waheed, 18. Rosa, R., Prdhu, M. O., Beuzard, Y. & Rosa, J. (1978) J. Clin. A., Gottschalk, S., Grzeschik, K.-H., Hasilik, A. & Figura, Invest. 62, 907-915. K. V. (1988) EMBO J. 7, 2343-2350. 19. Rose, Z. B. & Liebowitz, J. (1970) J. Biol. Chem. 245, 3232- 4. Van Etten, R. L., Davidson, R., Stevis, P. E., MacArthur, H. 3241. & Moore, D. L. (1991) J. Biol. Chem. 266, 2313-2319. 20. Lowry, 0. H., Rosebrough, N. J., Farr, A. L. & Randall, R. J. 5. Rapoport, S. & Luebering, J. J. (1950) J. Biol. Chem. 183, (1951) J. Biol. Chem. 193, 265-268. 507-516. 21. Laemmli, U. K. (1970) Nature (London) 227, 680-685. 6. Rosa, R., Audit, I. & Rosa, J. (1975) Biochimie 57, 1059-1063. 22. Rosa, R., Prehu, M.-O., Albrecht-Ellmer, K. & Calvin, M.-C. 7. Fothergill-Gilmore, L. A. & Watson, H. C. (1989) Adv. En- (1983) Biochim. Biophys. Acta 742, 243-249. zymol. 62, 227-313. 23. White, M. F. & Fothergill-Gilmore, L. A. (1992) Eur. J. Bio- 8. Winn, S. I., Watson, H. C., Harkins, R. W. & Fothergill, chem. 207, 709-714. L. A. (1981) Philos. Trans. R. Soc. London Ser. B 293,121-130. 24. Poillon, W. N. & Kim, B. C. (1990) Blood 76, 1028-1036. 9. Cherfils, J., Rosa, R., Garel, M. C., Calvin, M. C., Rosa, J. & 25. Greaves, D. R., Fraser, P., Vidal, M. A., Hedges, M. T., Janin, J. (1991) J. Mol. Biol. 218, 269-270. Ropers, D., Luzzatto, L. & Grosveld, F. (1990) Nature (Lon- 10. Craescu, C. T., Schaad, O., Garel, M. C., Rosa, R. & Edel- don) 343, 183-185. stein, S. (1992) Biochimie 74, 519-526. 26. Ryan, T. M., Townes, T. M., Reilly, M. P., Asakura, T., 11. Dubart, A., Romeo, P. H., Tsapis, A., Goossens, M., Rosa, R. Palmiter, R. D., Brinster, R. L. & Behringer, R. R. (1990) & Rosa, J. (1984) Biochem. Biophys. Res. Commun. 120, Science 247, 566-568. 441-447. 27. Rubin, E. M., Witkowska, H. E., Spandler, E., Gurtin, P., 12. Taylor, J. W., Ott, J. & Eckstein, F. (1985) Nucleic Acids Res. Lubin, B. H., Mohandas, N. & Clift, S. M. (1991) J. Clin. 13, 8765-8785. Invest. 87, 639-647. 13. Sanger, F., Nicklen, S. & Coulson, A. R. (1977) Proc. Natl. 28. Trudel, M., Saadane, N., Garel, M. C., Bardakdjian-Michau, Acad. Sci. USA 74, 5463-5467. J., Blouquit, Y., Guerquin-Kern, J. L., Rouyer-Fessard, P., 14. Garel, M. C., Joulin, V., Le Boulch, P., Calvin, M. C., Prehu, Vidaud, D., Pachnis, A., Romeo, P. H., Beuzard, Y. & Cos- M. O., Arous, N., Longin, R., Rosa, R., Rosa, J. & Cohen- tantini, F. (1991) EMBO J. 10, 3157-3165. Solal, M. (1989) J. Biol. Chem.' 264, 18966-18972. 29. Ingram, V. M. (1956) Nature (London) 178, 792-794. 15. Maniatis, T., Fritsch, E. F. & Sambrook, J. (1982) Molecular 30. Wishner, B. C., Ward, K. B., Lattman, E. E. & Love, W. E. Cloning: A Laboratory Manual (Cold Spring Harbor Lab. (1975) J. Mol. Biol. 98, 179-194. Press, Plainview, NY). 31. Dykes, W., Crepeau, R. H. & Edelstein, S. J. (1979) J. Mol. 16. Calvin, M. C., Blouquit, Y., Garel, M. C., Prehu, M. O., Biol. 130, 451-472. Cohen-Solal, M., Rosa, J. & Rosa, R. (1990) Biochimie 72, 32. Eaton, W. A. & Hofrichter, J. (1987) Blood 70, 1245-1266. 337-343. 33. Luzzatto, L. & Goodfellow, P. (1989) Nature (London) 337, 17. Garel, M.-C., Lemarchandel, V., Calvin, M.-C., Arous, N., 17-18. Downloaded by guest on October 2, 2021